Click here for linear version of these pages.
  PT flag DE flag
About Us
New Account
Access Your Account
Info & Links
ANN Documentation
Data Security
User License Agreement
How to quote
Scientific Publications Related to the microCortex Algorithm
Why ANN?
To Get Started
WalkThrough Guide
Guide: Credit Risk Assessment
Guide: How to Use and Features
Guide: How to Analyse the ANN
Data Types
Very Quick Guide to Neural Networks
ANN: More Details logo
Privacy Statement
Find at our Site
Contact us

Details on ANN training

Artificial learning by ANN is tricky and computationally intensive. There are two major pitfalls that must be avoided: over-training and sub-optimal topology.

  1. Over-training leads to memorisation instead of learning. To avoid this, one needs to play a show/hide game with the ANN to make sure the predictions can be generalised (by bootstrapped early stopping with cross-validation).
  2. Optimal topology occurs when the size of the ANN matches the complexity of the problem. This is achieved by adjusting the complexity of the network itself during the learning process has developed and implemented the necessary algorithms to circumvent these common shortcomings of neural computing. By doing so, our application is able to identify a solution for the unknown dependency that is your best bet for a prediction. The theoretical remedies have been known for a while but its full implementation is computationally intensive and algorithmically challenging. We have automated the procedure and have implemented it in high performance computing set as an application provider.

To Get Started